Assembly Methods

ipyrad has four methods for assembling RAD-seq data sets. The first and simplest is denovo, which requires no prior information or genomic resources, while the remaining three methods all require some form of reference sequence. It is important to note, however, that many types of genomic resources can be used as a reference sequence, not just complete nuclear genomes. For example, plastome and transcriptome data can be used to partition reads among assemblies with different types of data, and reference sequences can even represent the genomes of symbiotic partners, or contaminants to be filtered/removed from a data set.


Sequences are assembled without any reference resources. Homology is inferred during alignment clustering by sequence similarity using the program vsearch.


Sequences are mapped to a reference genome using the program bwa_ (or smalt, optionally) based on sequence similarity.


Sequences are mapped to a reference genome based on sequence similarity, and reads that do not match to the reference are assembled using the denovo method.


Sequences which map to a reference genome are excluded, and all remaining reads are assembled using the denovo method. This method can be used to filter out data, such as reads matching to a chloroplast genome in plants, or to a host genome in a study of a parasite.

Combining multiple methods

You could imagine that if you had a reference sequence file you might want to examine your data set under a number of different Assembly scenarios. For example, let’s imagine we are interested in inferring phylogeny for a clade of 10 plant species and we download transcriptome data for a close relative of our focal clade. We could assemble our RAD-seq data set using only the data that match to the transcriptome (putatively coding regions), and compare this with results when we assemble all of the data that do not match to the transcriptome (putatively non-coding).

Example CLI combining assembly methods

## create a params.txt file and name it "coding". Then use a text editor
## to edit the parameter settings in coding-params.txt and enter the path
## to the transcriptome.fasta file for the 'reference_sequence_path', and
## enter 'reference' for the 'assembly_method' parameter.
ipyrad -n coding

## run steps 1-2 using the settings in data1-params.txt
ipyrad -p params-coding.txt -s 12

## create a branch called "noncoding" and edit the newly created file
## noncoding-params.txt. Set the assembly_method to 'denovo-reference'
## and leave the transcriptome.fasta file as the 'reference_sequence_path'
ipyrad -p params-coding.txt -b noncoding

## now run steps 3-7 for both assemblies
ipyrad -p params-coding.txt -s 34567
ipyrad -p params-noncoding.txt -s 34567

Example Python API combining assembly methods

## import ipyrad
import ipyrad as ip

## create an Assembly and modify some parameter settings
data1 = ip.Assembly("coding")
data1.set_params("project_dir", "example")
data1.set_params("sorted_fastq_path", "data/*.fastq")
data1.set_params("reference_sequence_path", "transcriptome.fa")
data1.set_params("assembly_method", "reference")

## run steps 1-2"12")

## create branch named 'noncoding' which inherits params from 'coding'
## and set the assembly method to denovo-reference so that it removes the
## reference matched reads.
data2 = data1.branch("noncoding")
data2.set_params("assembly_method", "denovo-reference")

## finish both assemblies"34567")"34567")

## See ipyrad analysis tools for comparing the results.